Literary Fields Forever Playing with the Book of Disquiet

The LdoD Archive: Collaborative Archive of the Book of Disquiet was published on the web in December 2017. Over the past four years we have organized many workshops in different settings, aimed at encouraging appropriations of the reading, editing


Beautiful and Useless
And I offer you this book because I know it is beautiful and useless.
-Book of Disquiet, "Perystile", c. 1913, transl. Zenith We are used to describing the literary field as a communication space with a rigidly defined structure.Such structure is characterized by the constellation of a set of functions (such as writing, editing, and reading) and actors (writer, editor, and reader) around a privileged artifact (the book).This conceptual rigidity thus extends to the very notion of the book, which is predominantly imagined as a closed and self-sufficient object in which textuality and materiality (either in print or digital media) coincide.What would happen if, instead of conceiving those three functions and the artifact that we call the book as predetermined and self-evident, we imagined them as open and dynamic processes?What would happen if we looked at them as manifestations of literary performativity, that is, of the set of actions that institute and maintain the very possibility of literary experience?
These two questions lie at the core of the LdoD Archive: Collaborative Digital Archive of the Book of Disquiet (2017Disquiet ( -2022)), a project of the Centre for Portuguese Literature at the University of Coimbra, coordinated by Manuel Portela and António Rito Silva.In the LdoD Archive, Fernando Pessoa's Book of Disquiet was transformed into a computer-readable and writable text whose machine readability and writability is itself a generator of multiple literary acts that allow us to observe and experience the dynamics that establish a literary field.
Thus, the LdoD Archive gives us the possibility of observing actual writing, editing, and reading processes whose conceptual and material horizon is the production of the occurrence "Book of Disquiet", that is, the historical instantiation of the authorial and editorial projectuality that resulted in a specific book-work.At the same time, the LdoD Archive allows us to experience those processes (reading, editing, writing) as a fluid game of literary roles that we can continue to play in the platform, motivated by the procedural nature of the book as an imaginary operator in each act of meaning construction (Portela 2017b(Portela , 2019;;Portela and Magalhães 2020).
In other words, this machine is an experimental textual environment for asking the following questions: What is an act of writing?What is an act of editing?What is an act of reading?What is a book?As an unfinished, fragmentary, and modular set of unordered texts in various stages of composition, the Book of Disquiet turned out to be the ideal work for modelling and simulating literary performance (Portela 2021b(Portela , 2022)).The LdoD Archive is a computational invention and creative space that allows us to observe and experience the procedural nature of the book and the acts of writing, editing, and reading in the second degree: a book in the process of becoming a book, writing as writing in process, editing as editing in process, and reading as reading in process.This literary machine is trying to approach those acts of becoming through the play of creativity.

Reading through the Book of Disquiet for the Nth Time
The temporality of perception of the string of writing is cognitively embodied in this system of instructions for reading.Units of writing -such as letters, words, sentences, paragraphs, sections, chapters, etc. -and units of printing -such as leaves, pages, quires, bound volumes, title-pages, covers, etc. -become entangled in the specific bibliographic form of the codex book.However, as random-access memories, codex structures continue to provide multiple entry points to their assemblage of symbolic marks.Their field of signs remains open for all sorts of stochastic motions through bibliographic space.Within the constraints created by folded and bound leaves it is possible to multiply our reading paths according to various protocols.
The multidimensionality of reading is also a function of the emergent relations created by this conjunction of written language, typographic form, and reading motions in the spacetime of the book.
The remediation of the bibliographic structures of the Book of Disquiet in the LdoD Archive multiplies the randomness of access.Firstly, the sequence of four editions is presented according to vertical and horizontal axes that allow for movements within each edition and across multiple editions.Each text from the Book becomes a traversable item according to specific editions or according to reader-constructed reading paths.Secondly, this construction of sequences also uses a topic modelling algorithm to produce an on-the-fly reading sequence that is responsive to reading as an ongoing process (see Fig. 1).In this instance, readers can follow a sequence of texts recommended by algorithmic analysis of the relation between texts already read and the entire set of still unread texts for a particular edition of the Book of Disquiet in the database.Thirdly, through the "Visual Book" interface readers can follow reading sequences on virtual editions according to their taxonomies or other metadata criteria.Readers can move smoothly from macro-visualizations of those variable structures and the actual textual blocks and strings (see Fig. 2).The aggregate result of those digital reading strategies is the transformation of codexbased random access into algorithmic random access.The radial and labyrinthine constellation of a very large number of possibilities becomes ergonomically manageable, bringing conceptualization and technical implementation into sync with each other.
The reading interfaces shown in Figures 1 and 2 illustrate our attempt to explore the modularity of the Book of Disquiet and of the digital medium to expand the reading motions through the writing space.In Figure 1, readers can navigate within a single editorial sequence or move across each of the four sequences.Additionally, in the recommendation column on the right hand side they can follow an algorithmic choice that is comparing the currently selected text with all the texts in the database not yet read.Predefined sequences by the expert editors, exploratory sequences created by readers, and variable automated sequences that respond to ongoing choices are presented next to each other.Reading complexity is increased, yet the cognitive plasticity modelled in the interface makes the labyrinth understandable and traversable without disorientation.The screen captures in Figure 2 are taken from the "Visual Book" interface, an application designed to provide a smooth transition between highlevel diagrammatic visualizations based on metadata and an uncluttered immersive reading surface focused on blocks of text.The first image shows a word-cloud representing the taxonomy in the "Jacinto do Prado Coelho annotated edition".If we click on the category "Língua-Linguagem" [Language] we are taken to the grid of squares in the second image: this grid of squares represents the whole edition (each square stands for one text), while the highlighted yellow squares correspond to those texts that have been annotated with the selected category.Once we mouse-over one of the yellow squares, the title of the text is shown.The third image presents the text reading screen.The angled brackets provide two distinct trajectories of reading: while the black arrow buttons allow you to move sequentially within the entire edition, the yellow arrow buttons take you sequentially to the previous or following text annotated with the selected category.The top menu contains other buttons of the "Visual Book" interface.After three seconds, all navigation aids disappear and only text and title remain, as shown in the fourth image.We have described this interface as an attempt to model the relation between reading as an immersion in the symbolic world of language and reading as a self-conscious exploration of navigation spaces (Raposo et al. 2021).

Editing through the Book of Disquiet for the Nth Time
That piece belongs to a book of mine for which I've written other, still unpublished passages, but I have a long way to go before finishing it.The book is called The Book of Disquiet, since restlessness and uncertainty are the dominant note.
-Book of Disquiet, "Excerpts from Letters", Letter to Armando Cortes-Rodrigues, September 2, 1914, transl.Zenith When working with the virtual editing functionalities, virtual editors can use the units established by expert editors as building blocks of user-generated gatherings.They can also annotate their editions through comments and taxonomic tags.A virtual edition can be defined as any aggregation of texts selected, ordered, and annotated by users of the LdoD Archive.Virtual editions of the Book of Disquiet thus instantiate one dimension of the editing process: the selection, ordering, and annotation of textual elements according to some rationale.
Even if the levels of document transcription and definition of basic textual units (such as paragraphs or texts) are not open to virtual re-editing (as interactors can only select from among the several versions for each text provided by the platform), the construction of different kinds of editions provides a powerful simulation of the editorial process and of its interpretative nature.As happens in the reading interactions described above, editorial processes also range from the human to the computational with variable degrees of distributed agency: they can be based on expert editors (by importing predefined arrangements of specific editions), on virtual editor decisions, or on algorithmically suggested choices.The following virtual editions highlight creative appropriations encouraged by this procedural and social approach to the Book of Disquiet as an edition in progress.In 2018, Nuno Meireles performed a staged reading of 130 fragments from the Book of Disquiet (see Fig. 3, above).This performance consisted of a constraint-based selection of texts using the affordances of the LdoD platform.The Virtual Edition "130 BrevFragMaqui" is described as follows:

Textual Editing for Performance
Edition created through successive recommendations, by the algorithm, from Pizarro's edition.Upon this machinic selection, the shortest 130 fragments were chosen without changing the reading order defined by the topic modelling algorithm.This virtual edition of 130 brief machine-selected fragments was created for the performance "Pessoa reads the Algorithm: A reading performance by Pessoa himself of 130 machine-selected fragments from the LdoD Archive".This reading, on June 13, 2018, at the School of Arts and Humanities, Amphitheater V, marked the 130th birthday of Fernando Pessoa (born on June 13, 1888) and the 180 days of the LdoD Archive: Collaborative Digital Archive of the Book of Disquiet (published on December 14, 2017).1 (Meireles, 2018; my translation) The criteria defined for this virtual edition showed the intersection between shorter texts that are part of the agreed corpus of the Book of Disquiet and many fragmentary pieces that are usually included as appendices.This unlikely reading sequence highlighted uncertainties of the writing process.At the same time it revealed how topics reappear in distant fragments, thus suggesting that expert editors were guided by semantic constellations for assigning or relating many brief texts to the Book of Disquiet project.The numerical constraint of fitting 130 pieces in 75 minutes of live performance contributed to this unique editorial experience of the text.

Textual Editing for Video
In 2021, the student theatre group Baile Verde from the University of Salamanca, Spain, published a series of videos staging the feminist performance "Sim, esta sou eu, Bernardo Soares" ["Yes, this is me (she/her), Bernardo Soares"], based on a montage of 28 fragments from the Book of Disquiet (see Fig. 4, below).The resulting virtual edition is presented as follows: This virtual edition is a record of the work of scripting the feminist performance "Sim, esta sou eu, Bernardo Soares" released on the Instagram profile "sim_esta_sou_eu_bs" on May 5, 2021, International Portuguese Language Day.The chosen fragments were collected from Richard Zenith's edition.They were sequentially organized and edited in order to produce a dialogue of three female voices that respond, from within the text, to the misogynist Soares.The excerpts used are annotated with the names of their respective actors, rehearsal guidelines, and links to the final video result. 2 (Magalhães et. al., 2021;my translation) By splitting the narrative voices of those 28 fragments into 45 scenes, the script focuses on the gendered nature of the speaking subjects in the Book.Highly negative descriptions of women as the "dirty sex" are reframed through a quartet of voices: Eduarda (misogynist), Celia (artistic), Inés (restless), and Cristina (reactive).Given that Pessoa's writing method was often based on the emulation and invention of discursive structures that enabled his heteronyms to write and think in particular ways, the transgendering and fragmentation of the sexist rhetoric of male voices in the Book is an embodied mode of textual critique.An instance of the virtual editing function of the platform, "Yes, this is me (she/her), Bernardo Soares" is also an original appropriation of the LdoD Archive as a creative textual environment.

Textual Editing for Film
In 2021, a virtual edition based on João Botelho's Film of Disquiet (2010) (see Fig. 5, above) was published in the LdoD Archive.This edition follows a reverse 2. See https://ldod.uc.pt/edition/acronym/LdoD-BaileVerde.engineering strategy which infers the film script from the text said or performed in the film.Given the director's characteristic practice of using verbatim words from the literary works that he uses as the basis for most of his films, this virtual edition shows the film as an edition of the Book of Disquiet that is based on a combinatorics of the internal and external modularity of its fragments.The stream of consciousness of the book's narrator is expressed through textual editing, transforming the film into a mixed-media work, that is, a filmbook.The "Film of Disquiet" virtual edition is described as follows: This virtual edition reconstructs the montage of the Book of Disquiet made by the director João Botelho for his Film of Disquiet (2010).The virtual edition "LdoD-Filme" is based on the edition by Richard Zenith.The ordering of the texts in this virtual edition follows the order in which they are cited in the film.In each fragment from the Book of Disquiet all passages used in all scenes of the film were marked.The scenes were numbered and annotated using a taxonomy that summarily identifies places, characters, and references in both textual passages and film sequences.A few interpolations that do not come from the Book of Disquiet were tagged as "Additional Text".The timing corresponding to the quoted passages was also marked (in the category field, this marking appears in the form "hourminute-second", for example, 00-23-38).In this way, it becomes possible to cross-reference both the various texts, the categories that group them, and the multiple sequences of the film, showing the complex and profound reinvention of the Book that takes place in the Film.The virtual edition "LdoD-Filme" aims to contribute to the understanding of the relationship between literary montage and film montage in João Botelho's work. 3 (Portela, 2021; my translation)

Textual Editing for Machine Reading
"LdoD-Mallet" is an edition with an algorithmic taxonomy created from the corpus of Pizarro's edition. 4The order of texts follows Pizzaro's edition.
Texts have been automatically annotated by means of a taxonomy based 3. See https://ldod.uc.pt/edition/acronym/LdoD-Filme.4. MALLET is the acronym for "MAchine Learning for LanguagE Toolkit", developed at the University of Massachusetts Amherst; see http://mallet.cs.umass.edu.
on a numerical weighting of terms according to TF-IDF (Term Frequency-Inverse Document Frequency) metrics.The topic modelling process for indexing this edition is explained as follows: Edition with automatically generated taxonomy for 30 categories, on the corpus of Jerónimo Pizarro's edition, using the Mallet topic generation software (which is integrated in the LdoD Archive), naming each category with the 3 most relevant words of the generated topic, associating fragments to the category if the percentage is greater than 11% and running 1500 iterations to generate the topics. 5(Portela 2017a, my translation) Automatic tagging is based on statistical analysis of word frequency weighted with the inverse ratio between frequency in the document and frequency in the corpus in order to determine three-word topics (excluding stop words) that are more specific of certain texts.Correlations show strong intersections of a limited number of semantic fields, and a significant statistical presence or co-presence of certain words: for instance "life", "being", and "soul" are associated to 604 texts, suggesting that they provide an underlying texture for other topics.The generated thirty-topic taxonomy of the Mallet edition is as follows:

Textual Editing for Games
6.This three-word topic in English is a consequence of the fact that the corpus includes a few fragmentary pieces with phrases, sentences, or annotations in English (generally presented in the appendices to the Book).Since the algorithm has been optimized for Portuguese only, these stop words in English were not removed.This category should be disregarded.The process of abstracting topics and finding semantic connections between paragraphs and between texts is the basis for the "LdoD Classification Game", a multiplayer online game developed for classifying texts from the Book of Disquiet (Marques et al. 2020) (see Fig. 6, above).The game is focused on reading paragraphs, assigning tags (open taxonomic categories) to those paragraphs, and then voting on the best tag for the whole text.Reading time is proportional to the size of each textual block, while tagging time and most voted categories are used for determining the scores of players at the end of each game.
The game is based on a virtual edition that uses the corpus of Zenith's edition.The taxonomy of this edition reflects the accumulated results of the multiple iterations of the classification game.A new game is automatically generated every 15 minutes.

Instructions:
• Each game requires a minimum of two players; • Games take place at a specific time to be defined when a new game is created; • The duration of a game depends on the size of a fragment (generally a game takes no more than 5 to 10 minutes); • Each game consists of 3 rounds, with rounds 1 and 2 occurring multiple times (one for each paragraph of the fragment): • Round 1 -Submit: The player has the first paragraph on the screen, reads and analyzes it, and then submits one and only one category that she considers appropriate (we advance to round 2 at the end of time); • Round 2 -Vote: The timer is now fixed at 15 seconds, the user sees the same paragraph that she saw in Round 1 and she also sees the categories submitted by all the participants.She can vote for only one category she considers appropriate (if there are no more paragraphs to analyze, then we advance to Round 3; otherwise, we go through Round 1 again, now to the following paragraph); • Round 3 -Review: The timer is set for 30 seconds, the user has the complete fragment on the screen, and at the top she can see the most voted categories so far.The user votes for a category for the entire text.She can change votes while time is available (but changing votes penalizes her points), and simultaneously she sees the total votes, that is, she is able to check in real time which category has the most votes, and thus she can see the variations in the number as participants vote.The aim is to determine the most suitable category to classify the fragment.
• At the end of round 3, the player who suggested the winning category is credited as the author of the tag in the virtual edition and corresponding fragment. 7(Marques, 2018;my translation) This gamification of reading may be useful in an educational setting: first, as a form of encouraging discussions around the identification of topics in texts, thus providing training in the practices of interpretation; then, as a practical experiment on text indexing and classification, which has implications for a fuller understanding of the role of metadata in structuring and retrieving information, particularly in digital environments.The collaborative creation of taxonomies will thus serve to demonstrate the problems of classification and the multiple perspectives created by different taxonomies and systems.

Textual Editing for Social Media
Twitter citations (in Portuguese) from the Book of Disquiet are automatically harvested (see Fig. 7, above).Since March 2018 more than 8200 quotations have been collected (Oliveira et al. 2018). 8Their integra-7.See https://ldod.uc.pt/classificationGames. 8. See https://ldod.uc.pt/citations.tion into the LdoD Archive is made, at one level, as an instance of social reading.Quotations are transcribed and automatically linked both to the source texts from the Book of Disquiet within the archive's database and to the external tweets in their respective accounts.At another level, this collection becomes a source for a social editing experiment: all fragments quoted during the last thirty days provide the content for the automated "Twitter Citations" virtual edition (Oliveira et al. 2019). 9In this virtual edition, texts are sequenced according to frequency of citation, in decreasing order.Thus the edition changes continuously in response to the social dynamics of online sharing of Pessoa's text and the time-based affordances of the digital medium.An experiment in linking social reading to social editing, the "Twitter Citations" virtual edition provides yet another perspective on the circulation of texts on networked environments.

Writing through the Book of Disquiet for the Nth Time
We'll be able to create secondhand; we can imagine one poet writing in us in one way, while another poet will write in a different way.I, having refined this skill to a considerable degree, can write in countlessly different ways, all of them original.
-Book of Disquiet, "The Art of Effective Dreaming for Metaphysical Minds", c. 1914, transl. Zenith If writing can be used for depersonalization, that is, for multiplying subject positions through specific uses of language, we can model the act of writing in ways that enhance this depersonalization.By inviting readers to play as writers and take specific textual passages as seeds for writing variations and extensions, the LdoD Archive externalizes the writing process.It enhances the workings of language in the production of subjectivity.The incompleteness and uncertainty that are self-consciously referred to in the Book of Disquiet are now experienced through actual writing acts that insert themselves in the discourse fields of the work.This metaphysical depersonalization of the writing self can be further encouraged though algorithmic tools that rewrite textual fragments 9. See https://ldod.uc.pt/edition/acronym/LdoD-Twitter.
according to formalized principles of semantic, syntactic, visual, phonetic, and multimedia association.
Examples of the productivity of writing through the Book of Disquiet can be seen in texts produced for the "Disquiet Variations" workshops in 2018 and 2019, and in the electronic literature work Machines of Disquiet (2014-2018) (see Fig. 8). 10 Twenty machines using the Book of Disquiet as their database were developed between 2014 and 2018.Each machine uses different poetic conventions and specific JavaScript interactive tropes.The screen capture in Figure 8 is taken from one of the interactions with "Machine Nº 10", a visual text that seems to frustrate the process of reading by superposing the black and white areas in ways that mix foreground and background for each letter/word (in this instance "Sensibilidade" [Sensitivity] and "Liberdade" [Liberty]).The partial coincidence expresses the forces of attraction and repulsion between the semantic charges of the chosen concepts as well as the dialectics between hiding and revealing.The perceptual challenge to disentangle the words is a reminder both of the interpretative action on the textual tensions in the Book of Disquiet and of the writing possibilities in programmable media.

The Book of Disquiet as Metaphysical Spacetime
Right now I have so many fundamental thoughts, so many truly metaphysical things to say that I suddenly feel tired, and I've decided to write no more, think no more.I'll let the fever of saying put me to sleep instead, and with closed eyes I'll stroke, as if petting a cat, all that I might have said.
-Book of Disquiet, Text 27, c. 1930, transl. Zenith The LdoD Archive is an experimental textual environment for simulating literary performativity through actual acts of reading, editing, and writing.As a performative model that invites its subjects to engage in the collaborative exploration of its affordances and constraints, it opens itself up to the processes of social semiosis that create and sustain a literary field of textual interactions.Insofar as its starting point is the modelling of the Book of Disquiet as a work under construction, it can be argued that the form and idea of the book have also been virtualized, that is, turned into a dynamic tool that can take many different embodiments.The operation of the form and idea of the book on the imagination of readers, editors, and writers has been digitally modelled through this multiplication of possibilities.
As a force field of our desire for meaning and closure, the book becomes yet another function in a network of symbolic and material action that calls upon the existential "restlessness and uncertainty" of living and writing, and living through writing.To a certain extent, the LdoD Archive has transformed the Book of Disquiet into a metaphysical spacetime in which the actions of creating literary form through reading, editing, and writing can be iterated again and again.A distributed and socialized instantiation of multiplicities, the LdoD Archive rehearses our literary acts as processes of becoming through a choreography of inscriptions.In this living space for play, the digital medium is reinvented as a new kind of literary experience, significantly distinct from the prescribed roles, institutional constraints, and regulated imagination of the print medium.

Figure 2 .
Figure 2. Traversing the scales of reading in the "Visual Book" interface: from infovisualizations of textual structure to typographic visualization of textual strings.

Figure 3 .
Figure 3.A staged reading by Nuno Meireles based on a virtual edition, June 13, 2018.

Figure 6 .
Figure 6.Homepage for the "LdoD Classification Game", a software application developed by Gonçalo Montalvão Marques.

Figure 7 .
Figure 7.A tweet with a quotation from the Book of Disquiet, published on January 15, 2022, and automatically collected by the LdoD Archive (screen capture).

Figure 8 .
Figure 8. Screen capture from Machines of Disquiet, a software application and work of electronic literature authored by Luís Lucas Pereira.